There are two startup-ish things I do that are not websites, but are largely disseminated via the web:
a) my main job is represented at gmatdaily.wordpress.com. I sell study materials for the GMAT ...I worked in the industry for a few years, discovered a multitude of problems, and am slowly solving them all.
b) i collect, package, analyze, and sell college baseball statistics to major league teams ... my partner and I had a website for this (collegesplits.com) where we gave away a lot of the data for free, but eventually took it down because the maintenance wasn't worth it. My web presence in the field made it easy to line up customers, and I suppose I could set up a members-only site to deliver the data (and thus the business would be a website), but that's not what the clients want.
Now, perhaps one of these days I'll work on something that IS a website :).
So we have all of the pain of developing a web-based application (browser incompatibilities, limitations of the medium, etc.) with all of the negatives of installed applications (high barrier to adoption, lower volume, etc.). But it does give us a business model that everyone can easily understand: Give us money, we give you software, you install it, we support you. Next year, you give us money again and we keep supporting you.
Web applications are the most fun to build, but there's an awful lot of room for technology in non-web spaces. Large retailers a huge users of technology (and their websites are generally not even a blip on the radar, as far as technology expenditures go--I was involved briefly in a $2.7 million content distribution deployment for Lowe's...I'm certain they haven't spent more than a tenth that much on their website).
Lots of areas for high tech to make a huge impact on peoples lives (and thus make a huge impact on your bottom line): Medical records and billing, legal services, banking, accounting, warehouse automation, etc. There are businesses working in all of these spaces already, of course, but there's still plenty of niches left unfilled.
Seriously: Because we don't want to be evil. We're willing to not have rounded corners in some browsers, or have menus that don't get animation, or whatever...but we're not willing to prevent anyone (even folks who are blind or otherwise have accessibility problems) from using our product.
This is why I think mobile has such great potential. A platform that the user carries around with her all day that has rich sensors such as camera, microphone and NFC. It's perfect! (apart from the small screen but thats a small price to pay for such access to an user)
for example, here's an idea that's been kicking around in my head. (i'm not the right guy to start this company, so i'll just throw out this there.)
you know how people at the gym often track their workouts -- "3 sets, 10 reps, 150 lbs" -- with pencil and paper? that's so 20th century. why not build a sensor network for exercise equipment? you get on a scale or weight/cardio machine. it knows who you are -- RFID in your membership card? -- and it wirelessly uploads your workout data from that machine to a server.
as a gym member, you get a login to an accompanying web app, which is automagically populated with your data. you can set goals, track your workouts + results with pretty graphs, etc. no more paper+pencil tracking -- just work out and log in.
sell this sensor/software platform to gyms (or perhaps the equipment manufacturers), and you've got yourself a business.
where does the internet come in? well, aside from the tracking app for individuals, with all this aggregate data, you'd revolutionize exercise science. is interval training really better than marathon cardio sessions? more weight+fewer reps, or less weight+more reps? even basic collaborative filtering would be cool: "people with bodies like yours got these results with this workout plan."
it'd disrupt the (absurdly lucrative) personal training industry.
point is, out there in the "real world," there's tons of data that's just waiting to be aggregated and analyzed. that'll certainly be the basis of many interesting companies.
For another, I'm going to advertise a blog we already have that has a small but growing audience. We'll have a small inventory with a limited number of skus. We'll probably make the items ourselves. I'm not sure how I'll handle the commerce (ebay, yahoo store, paypay...). We'll do fulfillment ourselves too. Once we establish a microbrand I have what I think is a clever bundling idea to increase our average transaction size by about 2-5 fold without needing to bloat our inventory. At that point, I'm not sure exactly what's next, we might try it on a small scale, continuing doing our own manufacturing, or we might line up offshore manufacturing and outsourced fulfillment and make a big push. I think we'll probably do a blend of the two. Start rolling out the evolved business model while still making the items ourselves at the same time we're lining up manufacturing and fulfillment with the goal for having them on-line in time for the holiday shopping season.
I also have ideas for web apps, but I'm cautious about both ad revenue and hoping to be bought out. I do have an idea for a service that would be monetized by transaction based revenue, but I have quite figured out how to put it into practice.
Too bad, he's one of those guys who just likes to invents to see if it works. Not much of a 'start a company' guy...I think he just wants to sit back and enjoy the rest of his life as opposed to dealing with the politics of a company. 9 years on working on something and the fear of someone stealing your idea keeps you from sharing it to the world. It's a shame but oh well. It's his 'baby'
I have my own 'baby' project to work on.
Anything you do on the web has to funnel every piece of data through that web app because browsers cannot access anything other than their originating server (They might soon be able to do that via Flex/Silverlight type plugins). Even if they could, it just doesn't make sense for many data centric interactive applications to buy a lot of servers when a huge number of powerful PC clients does almost nothing.
I reckon, if I have some algorithmic data analysis stuff and I can distribute that to client machines to some degree, I can make my service cheaper and maybe even architecturally simpler in some cases, because there is a natural partitioning scheme. Ok, now I see this argument becomes quite obscure without going into further detail so I'll leave it at that.
For example, I read about someone who created a startup selling knife sheaths on EBay. Even if he were to create a website advertising his knife sheaths, he still wouldn't really have a webpage startup.
So, is anyone working on a non-webpage startup, and if so, what're you doing? :)
Shawn
I am working on software development tools: http://www.codesynthesis.com
I've done both desktop/server software and web software, and IMHO good web software is harder. With web software, you have to explicitly think about many things that just aren't an issue with conventional software, like how to distribute the load of a million users over dozens of boxes. It helps that user expectations for UI responsiveness and robustness are somewhat lower with webapps, but that's changing with AJAX. You also need to know many more technologies to build a successful webapp: CSS, HTML, Javascript, a web scripting language, SQL, database performance, server administration & shell scripts, and any libraries necessary for your problem domain.
It is a completely different skillset. With conventional apps, you need to know how to architect a large single app instead of architecting a large collection of collaborating servers. You need to know C++, Java, or C# instead of PHP/Python/Ruby. You need to know intimately the particular toolset that your app fits into - VMWare can't live without device drivers; Sleepycat can't live without transaction algorithms; one desktop app I worked on needed an intimate knowledge of TCP, including Winsock LSPs and Linux kernel hacking. You get much more out of deep knowledge with conventional apps, while webapps require a more shallow understanding of a very broad range of topics.
Also, the technologies you need for desktop apps don't have zillions of articles on them plastered across the web, and the documentation is often poor quality. But if you take a professional software engineering job, you'll be expected to make sense of it and figure out how everything works (possibly through prototypes, debuggers, and a lot of pain and patience). It's not taught in college, but that doesn't make it particularly difficult.
I am not saying web development is necessarily easier. I don't have any problem believing Javascript or AJAX is a lot of pain to get working consistently across various browsers. What I am saying is that serious software (i.e., the kind that people are willing to pay for) requires deeper knowledge and practical experience to get right.
Most profitable software requires deep knowledge of at least one relatively unknown technology (the three web startups mentioned above are exceptions). This applies as much to web startups as desktop startups, eg. Google had deep knowledge of search, YouTube of Flash and video, Amazon of inventory and retailing. But acquiring this deep knowledge is not rocket science: typically, you identify the technologies that'll be useful for your app, you read all the publicly available documentation, you download a demo and write some prototypes, you ask some questions on a mailing list, and you e-mail the support people at the vendor (possibly paying for a product along the way). Fresh college graduates typically don't have this knowledge off the bat because it's not taught in school and they aren't exposed to it, but that doesn't prevent them from picking it up quickly. Look at Loopt, for example.
I can only ask what kind of software have you developed, sir?
If web software was harder, you wouldn't see "products" created over a weekend shaping up into companies a week later. And don't get me started on "millions of users on dozen boxes".
The primitiveness of web application development is mainly the reason why so many "ex-taxi drivers" and are rolling out web apps daily.
Two huge reasons why web app development is so trivial are: - You control your hardware. This is really important. - You do not need to develop UI. Web Pages are joke compared to the "real thing".
For starters, try to develop an super-simple desktop application that uploads an arbitrary file to a given FTP address, let 100,000 users (let alone millions) download it to discover: - It won't start - It won't connect - It crashes their computer. - "Where did it go?" question after downloading.
You'll be amazed. You will realize that PCs of most people are just like jungle, populated with adware, multiple software firewalls running simultaneously, paranoid anti-virus packages, missing DLLs and weird security settings. Moreover, users will be brutal: they'll be turning power off in the middle of your disk writes, they will upgrade libraries you linked to without asking you first, they will install your executable on desktop and will be killing files you create, the list goes on.
Over 80% of developing desktop software consists of two things: - Fighting unexpected problems in a hostile environment of user's PC - Dealing with real, LIVE UI that is reflecting real-time changing objects, and is pleasant to use.
On top of that you're trying to do this using minimal amount of libraries, because you can't expect people to download 100MB installer.
But webapps have the same problem, they just push the complexity into a different area. I also volunteered for FictionAlley.org, a mid-size (110k or so registered users) web community. We would get complaints from users that the site was unreadable on their 40 character PDA screen. We'd get e-mails from blind people saying could we please make the site handicap-accessible and usable on their screen readers. We would get irate e-mails from Opera users about display glitches, or irate e-mails from Mac users wondering why the latest feature we just released didn't work on Safari. We'd get tirades from people because we added navigation bars, after getting tirades from people because we didn't have navigation bars. We'd get constant complaints about the (free) site being too slow, because we didn't have money for more servers.
Until you've actually been involved with a website that has lots of visitors, you have no idea what goes on behind the scenes to keep things running smoothly (or sometimes not-so-smoothly). Yes, desktop apps are hard. I've done them - not just the release engineering stuff above, but I also wrote the Java app included in the package, and I wrote a Netbeans plugin for my current employer, and both had to deal with the sticky platform-and-threading issues necessary for a responsive, professional UI. I still say that webapps are harder.
I've worked on two large projects that are distributed and not web based.
The difference is that in a webapp, you have to think of the implications of distribution on every user request. So you have to consider latency and status feedback for everything the user does, you can't use shared state, you have to keep in-memory data structures to a minimum, you can't count on getting notification when events occur, and so on. Desktop-based apps with a distributed back-end usually have a well-defined interface to the cluster. As long as you don't need to update the backend, you can keep anything you want in memory. And you usually have a two-way connection: you don't need to frame everything in terms of request/responses.
www.uuorld.com
We're using trolltech's Qt and releasing on Win32, OS X, and Linux. If anyone knows of any mature packages for interactive 3d over the web, I'm all ears.
"Mature packages for interactive 3D over the web"
They are general purpose game building frameworks. I am currently working with it in MMO form, all implementation code in Python. It is a mature package for interactive 3D over the web if I ever saw one.
Hmm.. there may be something to that.... hey thanks! :)
After talking to a couple people in biotech start ups, it seems pointless to pursue something unless I am a cofounder. The marriage of science and business is often not pretty.
Protocol is named DITP (distributed information transfer protocol) and uses IDR (information data representation) as encoding rule (it's binary). The infrastructure is named DIS (distributed information system).
Some usable code will be out in a few weeks. The system is original in many ways and the curently forseen business model might be too. Coming out will be announced on Y Combinator news.
Looking for ways to license visual search technology.
web - handhend - desktop - web
jobs@churchillnavigation.com
a startup that's not a website
1. Mobile search. No website. If we did it would be a one static page site.
2. Image recognition. Also no interaction with a website. As far as goods, I would stay away from it. Logistics r pain in the butt.